Does pattern-welding make Anglo-Saxon swords stronger?
Transcript of Does pattern-welding make Anglo-Saxon swords stronger?
1
Does pattern-welding make Anglo-Saxon swords stronger?
Thomas Birch
Abstract
The purpose of pattern-welding, used for the construction of some Anglo-Saxon swords, has
yet to be fully resolved. One suggestion is that the technique enhanced the mechanical
properties of a blade. Another explanation is that pattern-welding created a desired aesthetic
appearance. In order to assess whether the technique affects mechanical properties, this
experimental study compared pattern-welded and plain forged blanks in a series of material
tests. Specimens were subject to tensile, Charpy and Vickers diamond hardness testing. This
was to investigate the relative strength, ductility and toughness of pattern-welding. The results
were inconclusive, however the study revealed that the fracture performance of pattern-
welding may owe to its use. This paper arises from work conducted in 2006-7 as part of an
undergraduate dissertation under the supervision of Dr. Catherine Hills (Department of
Archaeology, University of Cambridge).
Introduction
Pattern-welding is the practice of forging strips (sheets and rods) of iron, sometimes of
different composition, that have often previously undertaken physical distortions such as
twisting and welding (Buchwald 2005, 282; Lang and Ager 1989, 85). It was mainly used in
the production of swords and spearheads. The purpose of pattern-welding, however, remains
debated. This study sought to understand the purpose of pattern-welding through experimental
investigation, with particular reference to why it was used to manufacture some Anglo-Saxon
swords. In order to establish whether pattern-welding improved the mechanical properties of a
sword, a sample was created and compared to a plain forged control sample. The standard
material testing methods employed to compare samples investigate strength, toughness and
hardness. Before presenting the methods and results of this experimental study, this paper
begins with a short review of the archaeological evidence, followed by a background to
academic research and discussion, of pattern-welding.
2
It is not within the scope of this paper to explore aesthetic arguments about pattern-welding
and so only one purpose is investigated here, the suggestion of improved mechanical
properties. It is important to realise that perceptions of 'strength' may be conceptually
different to modern notions, which often pertain to functional and physical properties. For
instance, a recent investigation into the distribution of Anglo-Saxon swords has highlighted
that perceived strength may be related to origins of manufacture (Birch 2011). The perception
of swords as animated objects of strength, as revealed in Anglo-Saxon depictions of swords in
art and literature, will be explored in a future paper.
The archaeological evidence for pattern-welding
One of the earliest examples of pattern-welding known from Britain is a blade fragment found
in lake Llynn Cerrig Bach (Anglesey, Wales), dated between the 2nd century BC and the mid-
1st century AD (McGrath 1968, 79). Other examples have been found in mainland Europe,
demonstrating that the technique has its origins in the Late Iron Age. Pattern-welding is more
commonly associated with swords and spearheads found in Northern Europe from the 2nd to
the 6th century AD, particularly the famous war booty sacrifices found in Southern
Scandinavia (Buchwald 2005, 264–291). Fourteen double-edged blades deposited at Vimose
(Denmark) were pattern-welded, dating to 210–260 AD and likely to be of Roman
manufacture (Davidson 1962, 32; Jensen 2003, 229–230; Maryon 1960a, 27). Ninety pattern-
welded blades have also been recovered from Nydam (Denmark). Contemporaneous with the
corpus found in Anglo-Saxon Britain, pattern-welded swords have also been found on the
continent in Frankish and Alemannic graves, and also in Latvia (Davidson 1962, 33; Tylecote
and Gilmour 1986, 253).
The technique of sword manufacture reached its peak during the 6th and 7th centuries AD and
is generally accepted to have ‘passed out’ by the end of the Viking period, declining notably
during the ninth century (Jones 2002, 145; Thålin-Bergman 1979, 122). It is unclear whether
the decline was due to fashion, or availability of better ores and steel (Davidson 1962, 32;
Lang 2009, 239; Tylecote and Gilmour 1986, 253). More recently, religion has been cited as
another factor in explaining the demise of pattern-welding (Gilmour 2010). Pattern-welding
did, however, continue into the 12th century in the manufacture of seaxes, mainly in
continental Europe (Tylecote and Gilmour 1986, 253).The technique was also used for
3
constructing samurai blades in Japan, particularly with the Gwassan school during the 16th
century, and was also used to produce the Malaysian kris of South-East Asia (Craddock 1995,
272–273; Davidson 1962, 35; Maryon 1960a, 35). Pattern-welding should not be confused
with damascening, which is distinctly different (Anstee and Biek 1961, 71; Craddock 1995,
275; Maryon 1960b, 52).
The purpose of pattern-welding: a history of archaeological interpretation and
experimentation
It was the experimental work of Maryon (1960a, 1960b) and Anstee and Biek (1961) that has
allowed us to understand how pattern-welded swords were made. Despite this achievement,
the purpose of the technique still remains to be clarified. Was the technique primarily
functional in that it improved the physical qualities of the sword? Or, was it intended to
enhance the decorative appearance of the object?
Initial studies by France-Lanord (1980) in 1948 and by Salin (1957) concluded that pattern-
welded swords were physically superior to ordinary blades because they were extremely hard,
highly flexible and very sharp. In 1961, Anstee and Biek's experiments concluded that the
pattern created by twisting the composite iron rods was merely a by-product of its
functionality (1961, 85). Tylecote also argued that the technique introduced carbon deeper
into the blade and increased its hardness (1962, 250). It seemed, therefore, that pattern-
welding was deemed to be a technique to improve the physical properties of an object.
However, Tylecote's opinions changed in later publications, and so did the overall discussion
on pattern-welding. He stated that it was unclear whether pattern-welding made weapons
appreciably stronger than simple piling (Tylecote 1976, 57), before later arguing that the
technique was more concerned with appearance. An investigation into edged weapons by
Tylecote and Gilmour concluded that pattern-welded swords were primarily designed as
ornamental or prestigious weapons and that the, “complexity of the patterns bore little relation
to the overall functional qualities of the blade” (1986, 251). They further argued that
phosphoric-iron was utilised to emphasis the patterned appearance of the blade rather than
improve its physical properties (1986, 249, 251). Tylecote concluded that pattern-welding was
principally employed for its decoration, whilst also acknowledging the technical
improvements that resulted from the technique.
4
Some Viking and Carolingian swords have a composite construction with thin layers of
pattern-welding welded to a core, like a veneer. This has led some scholars to believe that the
technique was mainly intended to be decorative (Ypey 1983), employed predominantly for its
aesthetic and symbolic qualities (Craddock 1995, 271–2; 2010; Leahy 2003, 123). The
suggestion that pattern-welding was primarily decorative has often been opposed to the
alternative explanation of improved function. More recently it has been suggested that while
one purpose was intended, the other may incidentally have been achieved.
As a result of their extensive X-radiographic study, Lang and Ager (1989) concluded that the
main purpose of pattern-welding was for its appearance. However, they acknowledged that,
“at present there is insufficient evidence to determine whether pattern-welding was primarily
for strengthening or decoration, but it is clear that the latter was important” (Lang and Ager
1989, 115). The purpose of pattern-welding remains elusive partly due to the absence of
conclusive evidence for either the aesthetic or functional argument.
This study arises from the state of knowledge as of 2006, where experimental research
seemed the logical way forward. Experimental work has become the main focus of more
recent studies into pattern-welding in a bid to resolve, if possible, the primary purpose of the
technique. Comprehensive work by Janet Lang (2007; 2009; 2011), brought to the attention of
the author during the preparation of this paper, has provided invaluable insights into the
technique. Her experiments into the mechanical properties of pattern-welding have
demonstrated that twisting of different irons improved some physical properties of the metal.
Practical experiments by Pelsmaeker (2010, 74) concluded that pattern-welding did provide
improved weapon performance over 'mono-steel' blades. The subjective nature of
Pelsmaeker's (2010) testing methods, however, prevents any objective comparison. The test
results presented in this study aim to complement the empirical work achieved so far.
The Anglo-Saxon sword: manufacturing the experimental samples
In order to prepare the samples necessary for testing the physical properties of pattern-
welding, careful consideration was taken in the selection of material, construction design, heat
treatment and surface treatment. Two samples were made for comparison. A pattern-welded
sample (PW) and a non-pattern-welded control sample (NPW). Whilst PW was constructed
from multiple units, the control sample NPW was simply forged from a single piece of iron.
5
The samples were manufactured by blacksmith Hector Cole at his forge in Little Somerford
(Wiltshire). Each sample was cut into three equal sized pieces that formed the blanks used for
sub-sampling to produce the specimens for mechanical testing. The aim of this experimental
approach was to best resemble pattern-welding of an Anglo-Sword construction, for testing,
as is evident in the archaeological record.
Material
The metallographic analysis of Anglo-Saxon swords and the Nydam blades has shown that
many were piled together from irons of different composition, with varying content of carbon,
phosphorous, as well as other alloying elements like nickel and sulphur. The Nydam blades
were constructed of high and low-carbon irons, whereas the Anglo-Saxon blades were
consistent in their use of low-carbon irons. The Anglo-Saxon swords, however, utilised high
and low-phosphorous irons, that may have been exploited to enhance the contrast in
appearance of the patterns on the blade. It may also have been utilised for its ability to harden
iron (Goodway 1999). Due to the low-carbon content, it may also have strengthened the iron
without embrittling it (Goodway and Fisher 1988, 22). It is accepted that the modern materials
adopted in this study may not be similar to those available to Anglo-Saxon smiths. However,
considering the variability in stock material used to construct pattern-welded swords in
antiquity, determining the right material to use becomes a practical impossibility. PW was
constructed from a low-carbon iron (0.2% C) and Victorian phosphoric wrought iron. NPW
was made from a single piece of low-carbon iron, the same used to make PW. Both PW and
NPW were made to the same dimensions.
Design and construction technique
Experimental work by Maryon (1960) and Anstee (1961) provided the final verdict on how
pattern-welding was achieved. Not only did they demonstrate a far simpler method than
earlier suggestions, but their method is now widely accepted as being correct. It involves the
twisting of multiple rods that are forged together into bundles, where several bundles are then
forged together to produce the final product. As a result, the earlier proposals for the pattern-
welding process were discredited. In 1948 France-Lanord attempted to construct pattern-
welded blades via a repeated folding technique. Another proposed construction by Janssens
(1958) involved the coiling and welding of a composite strip around a pentagonal core,
6
subsequently removing selected sections. The latter two suggestions were contested due to the
great deal of time and effort they required as well as their impracticality. Therefore, the
construction technique adopted in this study follows the process outlined by Anstee and Biek
(1961).
The design selected for constructing PW is shown in Figure 1, corresponding with pattern 'B1'
identified in Lang and Ager's (1989) study of Anglo-Saxon swords. Whilst a variety of
different patterns exist for Anglo-Saxon swords, the selected design appears to be one of the
earliest and most common patterns from the 5th to 7th century AD. Any original purpose of
pattern-welding may relate to the earlier designs, as opposed to the more complex patterns
and constructions that appear later.
Fig 1. A diagram showing the stage by stage process for the pattern-welded construction of
PW. The flat strip that is twisted with two rods (used to make a bundle) is highlighted in grey
in the smaller image accompanying the second stage. Three bundles are finally welded and
forged together.
Having examined the X-radiographs of the swords at the British Museum of the B1 design, it
was observed that the width of the bundles and the interval between twists varied greatly
within and between swords. The width of the bundles of the Hurbuck (Durham) sword varied
between 8–10mm, with some areas reaching a minimum width of 4mm. The interval between
twists varied between 4–7mm. The Faversham blade was much more consistent, where
intervals between twists were consistently 2–3mm. The thinner the flat strips, the closer
together they can be twisted, and thus the greater angle they make with the axis of the rod
during twisting (Maryon 1960a, 29). Due to the variability presented in the archaeological
record, no specifications were made for the interval measurements between twists, nor the
angle of the twists for constructing PW.
7
The construction of PW (Figure 1) was achieved by forging together three bundles, each
containing three rods (two round and one flat). The flat rod (strip) in combination with the
two round rods creates the tightest twist (compared to three round rods) and minimises the
gutters or grooves created during twisting, trapping less slag and reducing the likelihood of
unwelded seams. The bundles were twisted to their full tightness. The angle of the flat strips
to the rod was 45–46º. The three finished bundles (composite rods) were then welded and
forged together. The central bundle remained a consistent width of 7mm. The two outer
bundles reached a maximum of 10mm, due to their slight widening during hammering. As
already stated, PW and NPW were forged to the same dimensions. The purpose of this study
was to focus on the properties of pattern-welding and so this variable was isolated for
investigation. No edge material was adopted in constructing PW as this would qualify as
another variable and technical feature worthy of investigation in its own right. Any
investigations into replica pattern-welded swords with cutting edges will overall reflect sword
performance and not necessarily the properties of pattern-welding.
Heat treatment
The affect of heat treatment on the blade depends very much on the carbon content, whereby
increased amounts of carbon means the blade is more likely to be affected. The pattern-
welding technique introduces more carbon into the blade than simple case-hardening,
whereby the rods and strips are being carburised on their surfaces when introduced to the
oxidising part of the hearth. These carburised surfaces are subsequently twisted and interned
into the blade. Any heat treatment is more likely to affect PW due to the likelihood of its
increased carbon content.
Anstee and Biek (1961) compared the outer bundles in experiment number 8 before and after
being tempered and annealed. They demonstrated increased tensile strength and hardness
values after heat treatment. The archaeological evidence for heat treatment in Anglo-Saxon
swords, however, shows chronological variation as well as different types. Tylecote and
Gilmour's (1986) study revealed that most of the early Anglo-Saxon pattern-welded swords
were not quenched. Of the eighteen early Anglo-Saxon swords examined, only six show
evidence of heat treatment in the form of quenched cutting edges to achieve a greater edge-
hardness (Tylecote and Gimour 1986, 245). Two middle Anglo-Saxon swords examined were
quenched, most likely in water. Of the eleven swords dating from the ninth to the eleventh
8
century AD, five were heat treated. One was possibly quenched in oil, two in water, and
another two were tempered and annealed (ibid, 248). Due to the evidence for heat treatment in
some Anglo-Saxon swords, both PW and NPW were quenched in water upon being finished
at red heat (about 800–900ºC). They were not annealed.
Surface treatment
Most Anglo-Saxon pattern-welded swords do not show signs of being ground away and few
had fullers. It appears there were differences in surface treatment between the British swords
and continental types that go beyond the scope of this paper. No surface modification in the
form of grinding, polishing or etching took place on the samples. Test specimens were
prepared from PW and NPW in their as-finished condition.
Examination of the samples: X-radiography and metallography
Prior to the sub-sampling for the preparation of test specimens, PW and NPW blanks were
examined for any internal faults. This was achieved through X-radiography and an assessment
of the microstructure. Standard radiographic exposures of the samples (2–3mm mean
thickness) were obtained with a Pantak 160kV CP Unit using fine-grained Fuji 80/50 film,
with an accelerating current of 20mA and a 60s exposure time. The source dimension was
3mm with a focus film distance of 80cm and a 90º beam angle. The X-radiographs were
viewed on a light-box with a tube current of 90kV. Macro-section from PW and NPW were
etched using nital and examined to assess the microstructure using a metallographic
microscope in reflected light mode.
9
Fig 2. Photograph (top) and corresponding X-radiograph (bottom) of the six blanks used in
this study: first three blanks are from sample NPW and the last three blanks are from sample
PW. The design of PW can be seen in the accompanying close-up image of the surface of one
of the blanks (right).
The X-radiograph (Figure 2) showed that the weld seams in PW were successful throughout
the sample until the remaining c. 15mm of one end, where the weld terminates unsuccessfully
(cavities observed). A macro-section extracted from this area showed that the two weld
interfaces between the three composite rods were successful in the central portion of the blade
section, but that they became more partial and eventually incomplete towards both surfaces of
the sample (Figure 3).
10
Fig 3. The macro-section of sample PW (top), highlighting the localised failed weld seam with
its cavities identified through X-radiography; two distinct zones (bottom) defined by the
presence and absence of slag inclusions, with uniform recrystallised ferrite grains and an
uneven distribution of carbon (pearlitic structures) due to the different irons used in the
manufacturing process.
Varying grain sizes were observed in PW owing to the different irons used to manufacture the
sample. The macro-section of PW, seen in the montage of Figure 3, revealed that the sample
had one well defined region of unevenly distributed slag inclusions and a few areas of
oxidation along weld-line interfaces (Brick and Phillips 1949, 41; Nutting and Baker 1965,
121).
11
No faults were observed in NPW. The montage of NPW, seen in Figure 4, shows a uniform
structure of recrystallised ferrite grains with no pearlite observed. Some spherodised carbides
were observed. The recrystallisation of the ferrite grains, usually indicative of annealing,
shows here that the sample was not cooled rapidly. Oxides were observed in places along the
surface of the sample, indicating some oxidation took place as part of the forging and/or
quenching process.
Fig 4. The macro-section of sample NPW (top), showing uniform recrystallised ferrite grains
containing spherodised carbides (bottom).
Mechanical testing for the properties of pattern-welding
12
This section describes the three tests that were performed on PW and NPW to test for strength
and ductility, resistance to fracture and hardness. PW and NPW were cut into three equal
sized blanks that were sub-sampled to produce the necessary test specimens for tensile,
Charpy and Vickers micro-hardness testing. The examination of PW and NPW outlined in the
previous section proved invaluable in deciding the locations from which to extract test
specimens. A schematic diagram of the sub-samples removed for testing can be seen in Figure
5.
Fig 5. Schematic diagram illustrating the sub-sampled locations to produce the test
specimens. The three blanks for both PW and NPW were sampled as follows: one blank to
produce two Charpy specimens (red), two blanks to produce tensile specimens (blue), one
blank of which also produced the macro-section (green). Measurements are provided in
13
millimetres. An example of a sampled PW blank with accompanying images of prepared test
specimens can be seen in the top image.
Ultimate tensile strength testing
Two cylindrical tensile test specimens (Ø=3.55mm) were prepared from both PW (avoiding
the identified faults) and from NPW blanks in accordance with standard requirements. These
were subjected to standard ultimate tensile strength testing using an Extensometer number
100S/135 which was calibrated to BS EN ISO 9513:2002 class 1.0 requirements. The test for
ultimate tensile strength provides information about the strength and ductility of the material
being tested.
The test reveals the elastic limit of the material, whereby the strain (caused by the load) is
elastic. When the strain no longer returns to zero (as the load is increased) the specimen is
being plastically deformed permanently. A point is reached where the material continues to
plastically deform without an increase in the load. This is known as the yield point. The
tensile testing in this study will not measure the yield point, but instead will measure the proof
stress (the stress value for a small amount of permanent plastic strain). This is because some
steels do not exhibit a yield stress. The test will also determine the ultimate stress reached (the
maximum stress the material reaches), providing a measure of the ultimate tensile strength.
The behaviour of a specimen during testing is also informative about the materials ductility. A
brittle material will usually break and fail at the ultimate strength. A ductile material will
continue to stretch as the load is increased, plastically deforming in order to reduce the stress.
The elongation of the specimen (how much it has stretched) and its reduction in area
(minimum cross-sectional area) are measures of ductility. A ductile specimen will not only
stretch uniformly, but it will exhibit a trait known as 'necking'. Necking describes the non-
uniform deformation that occurs at a localised area in the specimen as the load is further
increased. It is also the point where the specimen will fail.
The Charpy test
Two test specimens (55mm x 10mm x 2.55mm, with 2mm V-shaped Charpy notch) were
prepared from both PW and NPW blanks in accordance with standard requirements. Two
14
Charpy tests were conducted at different temperatures; the first test at room temperature
(23ºC) and the second test at 5ºC. This was to assess whether differences in temperature may
affect the toughness of the specimens. The test specimens were subjected to 300 Joules
nominal striking energy and the machine calibrated to BS EN 10045-2:1993.The Charpy test
measures the amount of energy that can be absorbed by a test specimen on impact until it
fractures. A pendulum is swung to fracture the specimen on the side opposing the V-shaped
notch. The amount of energy required to fracture the specimen is measured (in Joules), thus
measuring impact toughness. It measures a material's impact resistance to fracturing.
The sub-sampling for one PW specimen included the unsuccessful weld-seam area previously
identified, as priority for specimen preparation was given to tensile specimens. Despite the
constraints of the study the affected sample was still deemed worthy of investigation.
Vickers diamond pyramid test
One macro-section was removed from both PW and NPW to produce a standard
metallographic polished block. Each section was prepared in a hot-setting Bakelite phenolic
resin using dialylphthalate glass fibre as a backing material to promote good edge retention,
and then flatly polished. A 10kg load was applied using a Vickers diamond pyramid indentor
that was calibrated to test specification BS EN ISO 6507-1:2005. Seven hardness values were
obtained for each macro-section (measuring the square-based diamond pyramid indentation
along the traverse of the specimen). The resistance to indentation provides a measurement of
the specimen's hardness and resistance to the load being applied.
Results
The results of the tensile, Charpy and Vickers diamond hardness testing of PW and NPW are
presented here. A discussion of the results follows this section with special attention paid to
fracture performance. The empirical data presented here should be regarded as informative
and complimentary to previously published studies.
Strength and ductility
15
The tensile testing results are presented in Table 1. Both NPW specimens fractured outside
the central third, as did one PW specimen, not in meeting with standard requirements (inside
the central third) for a satisfactory test. Fractures outside the central third usually give values
lower than those that would be obtained from within the central third. Therefore, the results
for the three unsatisfactory tests are likely to be lower than they should be. Despite only one
specimen fracturing successfully inside the central third, the results show differences between
PW and NPW specimens.
Specimen Proof stress
(N/mm2)
Maximum stress
(N/mm2)
Elongation
(%)
Reduction in area
(%)
NPW 1 331 422* 33.0 55.0
2 343 430* 33.0 64.0
PW 1 290 409 31.5 43.0
2 302 413* 25.5 24.0
Table 1. The tensile strength test results for two specimens extracted from NPW and the two
from PW. The asterisked results mark specimens that fractured outside the central third.
NPW specimens exhibited greater proof-stress and ultimate strength values than PW, and it is
likely that these values should be higher than those obtained. The maximum difference of 21
N/mm2 between NPW and PW should be considered marginal. Although NPW specimens are
consistent at 33% elongation, there is no immediate clear difference to the elongation values
obtained for PW specimens. The greatest difference recorded between NPW and PW
specimens is the reduction in area, where the maximum difference is up to 30%. The tensile
test results would indicate that NPW specimens are more ductile than PW, which is likely due
to the uniformity of the material, compared to the heterogeneity imposed by PW's composite
construction. One PW specimen displayed interesting features in the fracture surfaces, which
is addressed in more detail later on.
Impact toughness
The Charpy test results showed that both NPW specimens required more energy to fracture
than PW specimens (Table 2). At both room temperature and 5ºC, NPW specimens required
around 5 Joules more than their PW counterparts to fracture on impact testing. More energy
16
was required to fracture NPW and PW specimens at 5ºC than at room temperature, indicating
that they increased, if not maintained, their toughness at lower temperature conditions. NPW
specimens exhibited greater lateral expansion compared to PW. Lateral expansion increases
with toughness, and so the difference confirms the Charpy test results that the NPW
specimens were somewhat tougher than the PW specimens.
Toughness (Joules) Lateral expansion (mm)
23ºC 5ºC 23ºC 5ºC
NPW 19 20 1.29 1.40
PW 14 16 1.18 1.14
Table 2. The Charpy test results of NPW and PW specimens being impacted at room
temperature (23ºC) and at 5ºC.
The fracture performance of the Charpy test specimens is worthy of note. Both NPW
specimens fractured completely with smoother surface planes compared to the ragged tears
observed in the two PW specimens. One PW specimen (5ºC) failed to fracture completely and
remained partially intact. A more detailed discussion of the differences in fracture
performance will follow later.
Hardness
There is no significant difference between the mean Vickers hardness values obtained for PW
and NPW (Table 3), which can be confirmed by a two-sample T-test. The more interesting
observation is the variation in hardness values. Greater variation in hardness values can be
seen in PW with a maximum of 156 and a minimum of 108, reflecting the uneven carbon
distribution. The values across the macro-section of NPW are more consistent, confirming the
uniformity of the microstructure and absence of pearlitic structures. The reader may wish to
compare the hardness values of PW and NPW to other reconstructed pattern-welded samples
provided by Hector Cole that were also tested as part of this study (see Appendix 1).
Hardness (Vickers) Mean
NPW 136 142 138 133 133 121 128 133.00
PW 138 138 108 139 124 136 156 134.14
17
Table 3. The seven Vickers hardness values obtained traversing along the macro-sections of
PW and NPW samples, provided with the mean value.
Evaluation of the mechanical tests
There are many variables to consider when discussing these results. Before discussing the
qualitative results, a comment should be made on the test values obtained by tensile and
Charpy testing. In evaluating the experiment, two uncontrolled variables exist in the form of
grain size and inclusions. These were unintended and future experiments should seek to
promote material comparability.
Non-metallic inclusions often serve to reduce fatigue strength, and so the slag inclusions
present in PW (compared to the near absence of inclusions in NPW) is a likely explanation for
the difference in the values obtained. Similarly, the predominance of a smaller grain size in
NPW is also likely to have contributed to its greater performance. The influence of grain size
and slag inclusions may be inferred from the PW tensile specimen test results. Although one
PW specimen appears to have failed along a weld-line interface (Figure 6), it is unlikely that
this is the cause for the low tensile value (413 N/mm2) because the second PW specimen
(clean fracture) value is similar (409 N/mm2), pointing towards other causes (grain size and
inclusions).
The economic constraints of this study only allowed for a small sample size. Future
experimental investigations should employ a relevant sample size in order to subject results to
more rigorous statistical methods, lending greater weight to any conclusions being drawn. It is
fair to state that the information generated in this study is inconclusive. However, that does
not mean to say it is not informative, as will now be highlighted.
Discussion: the fracture performance of pattern-welding
The nominal values obtained by tensile and Charpy tests should be considered in conjunction
with the qualitative results. Differences observed in the fracture surfaces between specimens
may also lead to conclusions about fatigue behaviour.
18
A macroscopic examination of the fracture planes of the Charpy specimens revealed
differences between PW and NPW (Figure 6). In short, NPW specimens produced largely
smooth flat planes, a fracture characteristic of brittle materials. By comparison, however, PW
specimens produced rough undulating fracture plans more often associated with ductile
materials. As already described, one of the PW Charpy specimens failed to completely
fracture (Figure 6). This Charpy specimen remained partially intact, bridged by an unbroken
rod in the central bundle.
The composite construction of PW may affect the behaviour of crack propagation. The
'fibrous' nature of PW inhibits crack propagation because the rods, which can be considered as
'fibres', act behind the propagating crack. This is an extrinsic fracture toughening mechanism
known as contact shielding, where the rods absorb energy and encourage crack closure. As
observed in the aforementioned Charpy specimen, the crack terminated at the single unbroken
rod, along which the specimen remained intact. The shear lips on the second PW specimen
also infer this ductile quality.
Conclusion
NPW values for proof-stress and ultimate tensile strength (maximum stress) were marginally
greater than PW, indicating that NPW specimens were slightly stronger. The clearest
difference between NPW and PW in tensile testing was the reduction in area, where NPW
was deemed to be more ductile. Although both exhibited increased toughness at 5ºC, Charpy
testing confirmed NPW to have a greater toughness than PW specimens. The differences in
quantitative results are best explained by the observations made in the microstructures of
NPW and PW, owing mainly to differences in grain size and inclusions.
The qualitative differences observed in fracture performance between PW and NPW may
indicate one of the favourable properties of pattern-welding, and indeed its use in Anglo-
Saxon swords. The composite construction of a pattern-welded sword means it is more likely
to remain intact along one of its core rods than a uniform blade made from a single body of
iron. The fibrous quality promotes crack closure. The rods utilized to form a pattern-welded
construction may be likened to the thin veneer sheets used to form 'plywood'. Fracturing
plywood against the grain will not succeed in a clean, or necessarily complete, fracture. The
analogy between the composite rod construction of pattern-welding and 'plywood' (or 'fibres'),
19
helps to highlight one of the potentially important properties of pattern-welding. Pattern-
welding may have deliberately intended to achieve this type of fatigue behaviour, compared
to the clean and complete fracture that is achieved from a uniform microstructure. Keeping a
sword together under physical stresses no doubt would have been a desirable physical
property.
Fig 6. Images of specimens after mechanical testing: PW tensile specimen that fractured
along a weld-line interface (left); aerial and oblique views of the PW specimen that remained
intact along an unbroken rod (middle), and aerial and oblique views of the NPW specimen
(right).
Acknowledgements
This experimental study into pattern-welding would not have been possible without the
expertise and generosity of 'The Test House' (The Welding Institute, Great Abingdon), who's
director David Ellin offered his support and enthusiasm in organising sample preparation,
analysis and X-radiography. Special thanks also go to Nicholas Green, Dennis Fish, Peter
Robinson and Brian Kersey for their time, effort and helpful assistance. The commissioning
of the samples was kindly funded by Magdalene College (University of Cambridge), Dr.
David Beresford-Jones and my father David Birch, to whom I am sincerely grateful. Gwil
Owen and Dr. Colin Shell (Department of Archaeology, University of Cambridge) aided with
20
the photography and metallography respectively. Janet Ambers (British Museum) allowed me
to inspect the original X-radiographs from Barry Ager and Janet Lang's (1989) study. I would
also like to thank David Dungworth and Justine Bayley from English Heritage, for providing
me unprivileged access to the correspondence and study notes from John Anstee's original
experiments. The project also benefited from insightful discussions with archaeologist Mike
Anderton. Finding a blacksmith whose repertoire included authentic pattern-welding proved
extremely difficult. For this, I owe my utmost thanks to Hector Cole for making the samples
for this study. The author wishes to also thank the anonymous reviewers whose helpful
comments have guided the final outcome of this paper. All errors and omissions are my own.
This paper arises from work conducted in 2006-7 as part of an undergraduate dissertation
under the kind and helpful supervision of Dr. Catherine Hills (Department of Archaeology,
University of Cambridge).
Appendix 1
Below are the Vickers hardness values obtained from other pattern-welded sword replicas that
were tested as part of this study, manufactured and supplied by Hector Cole.
Hardness (Vickers) Mean
Brimble Hill Sword centre 170 157 159 176 174 167.2
edges 657 642 627 620 636.5
Saxon Sword Number 7 centre 113 143 142 138 151 137.4
edges 193 230 210 203 209.0
Sutton Hoo Mound 17
Sword
centre 119 131 148 151 145 138.8
edges 224 232 230 224 227.5
Roman Gladius centre 115 133 141 141 117 129.4
edges 185 199 192 178 188.5
Damascus Steel centre 209 194 197 203 188 186 183 185 193.1
Table showing the Vickers hardness values traversing along blade macro-sections, provided
with the mean hardness. All values were obtained from the mid-thickness. Edges examined
were always higher-carbon regions. Montages and micrographs of these sections are
available from the author.
21
References
Anstee J W and Biek L 1961, ‘A Study in Pattern-Welding’, Medieval Archaeology 5, 71–93.
Birch T 2011, ‘Living on the edge: making and moving iron from the ‘outside’, in Anglo-
Saxon England’, Landscape History 32, 5–25.
Brick R M and Phillips A 1949, Structure and Properties of Alloys (New York).
Buchwald V F 2005, Iron and Steel in Ancient Times (Copenhagen).
Craddock P T 1995, Early Metal Mining and Production (Edinburgh).
Davidson H R E 1962, The Sword in Anglo-Saxon England: Its Archaeology and Literature
(Woodbridge).
France-Lanord A 1980, Ancient Metals: Structure and Characteristics: Technical Cards =
Metaux Anciens: Structure et Caracteristiques fiches techniques (Rome).
Gilmour B 2010, 'Ethnic Identity and the Origins, Purpose and Occurrence of Pattern-Welded
Swords in Sixth-Century Kent: The Case of the Saltwood Cemetery', in M Henig and N
Ramsay (eds.), Intersections: The Archaeology and History of Christianity in England,
400-1200. Essays in honour of Martin Biddle and Birthe Kjølbye-Biddle (Oxford), 59-70.
Goodway M 1999, ‘The relation of hardness to strength in high-phosphorus iron wire’,
Historical Metallurgy 33, 104–105.
Goodway M and Fisher R M 1988, ‘Phosphorus in low carbon iron: Its beneficial properties’,
Historical Metallurgy 22, 21–23.
Janssens M 1958, ‘Essai de reconstitution d’un procédé de fabrication des lames d’épées
damassées’, Studies in Conservation 3, 93–106.
Jensen X P 2003, ‘The Vimose find’, in L Jørgensen, B Storgaard and L Gebauer Thomsen
(eds.), The Spoils of Victory: The North in the Shadow of the Roman Empire
(Copenhagen), 224–238.
Jones L 2002, ‘Blade construction and pattern-welding’, in I G Pierce (ed.), Swords of the
Viking Age (Woodbridge), 145–151.
Lang J 2011, 'The Celtic Sword: a source of practical inspiration', in H Jiří, H Cleere and L
Mihok (eds.), The Archaeometallurgy of Iron: Recent Developments in Archaeological and
Scientific Research (Prague), 263-318.
Lang J 2009, ‘A consideration of the methods of constructing iron sword blades in the pre-
medieval period’, in Archaeometallurgy in Europe 2007 (Milan), 232–240.
22
Lang J 2007, The Rise and Fall of Pattern Welding: an investigation into the construction of
pre-medieval sword blades. Unpublished Doctoral Thesis, University of Reading.
Lang J and Ager B 1989, ‘Swords of the Anglo-Saxon and Viking Periods in the British
Museum: A Radiographic Study’, in S C Hawkes (ed.), Weapons and Warfare in Anglo-
Saxon England (Oxford), 85–122.
Leahy K 2003, Anglo-Saxon Crafts (Stroud).
Maryon H 1960a, ‘Pattern-Welding and Damascening of sword-blades - Part 1: Pattern-
welding’, Studies in Conservation 5, 25–31.
Maryon H 1960b, ‘Pattern-Welding and Damascening of sword-blades - Part 2: The
Damascene process’, Studies in Conservation 5, 52–60.
McGrath J N 1968, ‘A preliminary report on the metallographic examination of four
fragmentary Early Iron Age sword blades from Llyn Cerrig Bach, Anglesey’, Historical
Metallurgy 2, 78–80.
Nutting J and Baker R G 1965, The Microstructure of Metals (London).
Pelsmaeker S B M 2010, Weapons of Princes, Weapons of War? An experimental analysis
from pattern-welded swords from northwestern Europe, 400-1100 AD. Unpublished
Master's Thesis, University of Groningen.
Salin E 1957, La civilisation merovingienne d’apres les sepultures, les textes et le laboratoire.
Troisiè Partie: Les Techniques (Paris).
Thålin-Bergman L 1979, ‘Blacksmithing in prehistoric Sweden’, in K Calissendorff and H
Clarke (eds.), Iron and Man in Prehistoric Sweden (Stockholm), 99–133.
Tylecote R F 1976, A History of Metallurgy (London).
Tylecote R F 1962, Metallurgy in Archaeology: A Prehistory of Metallurgy in the British Isles
(London).
Tylecote R F and Gilmour B J J 1986, The Metallography of Early Ferrous Edge Tools and
Edged Weapons (Oxford).
Ypey J 1983, ‘Frühmittelalterliche Waffen mit Damast’, in Damaszenerstahl: Vorträge der 1.
Fachtagung über "Damaszenerstahl — Stahlgewinnung und Stahlverarbeitung in der
vorindustriellen Zeit (Düsseldorf).
Author’s Details
Thomas Birch is a PhD candidate at the Department of Archaeology, University of Aberdeen.
Department of Archaeology
23
School of Geosciences
University of Aberdeen
St. Mary’s, Elphinstone Road
Aberdeen AB24 3UF
Scotland, UK